Increased Prevalence of TG and TPO Mutations in Sudanese Children With Congenital Hypothyroidism

Abstract

Context

Congenital hypothyroidism (CH) is due to dyshormonogenesis in 10% to 15% of subjects worldwide but accounts for 60% of CH cases in the Sudan.

Objective

To investigate the molecular basis of CH in Sudanese families.

Design

Clinical phenotype reporting and serum thyroid hormone measurements. Deoxyribonucelic acid extraction for whole-exome sequencing and Sanger sequencing.

Setting

University research center.

Patients

Twenty-six Sudanese families with CH.

Intervention

Clinical evaluation, thyroid function tests, genetic sequencing, and analysis. Our samples and information regarding samples from the literature were used to compare TG (thyroglobulin) and TPO (thyroid peroxidase) mutation rates in the Sudanese population with all populations.

Results

Mutations were found in dual-oxidase 1 (DUOX1), dual-oxidase 2 (DUOX2), iodotyrosine deiodinase (IYD), solute-carrier (SLC) 26A4, SLC26A7, SLC5A5, TG, and TPO genes. The molecular basis of the CH in 7 families remains unknown. TG mutations were significantly higher on average in the Sudanese population compared with the average number of TG mutations in other populations (P < 0.05).

Conclusions

All described mutations occur in domains important for protein structure and function, predicting the CH phenotype. Genotype prediction based on phenotype includes low or undetectable thyroglobulin levels for TG gene mutations and markedly higher thyroglobulin levels for TPO mutations. The reasons for higher incidence of TG gene mutations include gene length and possible positive genetic selection due to endemic iodine deficiency.

Congenital hypothyroidism (CH) is due to abnormal development or architecture of the thyroid gland (dysgenesis) or impaired thyroid hormone (TH) synthesis (dyshormonogenesis). Between 15% and 20% of worldwide cases with CH are caused by dyshormonogenesis (1). In the Sudanese population, the rate of dyshormonogenesis is 60% or 3 times that in the non-Sudanese populations (2). Several factors account for this disproportionate prevalence, including high rates of consanguinity and low levels of iodine intake. There are more than 60 known genes that influence thyroid development and TH function (Table 1). Dual-oxidase 1 (DUOX1) and dual-oxidase 2 (DUOX2) have homology in their structure with 7 transmembrane domains, and both are responsible for hydrogen peroxide generation, crucial for TH synthesis (3). Iodotyrosine deiodinase (IYD) acts to recycle iodide for future use in TH synthesis with its nitroreductase domain playing a critical role in the recycling (4). Several solute-carrier family member genes (SLC) including SLC26A4, SLC26A7, and SLC5A5 facilitate transport necessary for proper thyroid hormone synthesis. SLC26A4 and SLC26A7 are homologous, each with more than 10 transmembrane domains, and both are expressed highly in the thyroid; however, SLC26A4 is also expressed in the inner ear, and thus deleterious mutations present as deafness in the Pendred syndrome (5, 6). SLC5A5 is known to function as a sodium-iodide symporter, facilitating the uptake of iodide into thyrocytes needed for TH synthesis (4). Two of the most frequent gene defects causing dyshormonogenesis worldwide involve thyroglobulin (TG) and thyroid peroxidase (TPO). TG encodes thyroglobulin (TG), a large protein that serves in the synthesis of TH, its storage, and that of iodine as well (7). The structure of TG consists of 3 regions and a cholinesterase-like (ChEL) domain. All 3 regions assist in stabilizing the ChEL domain, which serves as a transporter for TG into areas of TH formation (8). Previously reported mutations of the TG gene showed a variety of phenotypes due to time of diagnosis, location of mutation within the molecule, and iodine intake (9).

Table 1.

List of Genes Related to Thyroid Disorders

Genes Related to Thyroid Disorder
AADAT	ALB	ALMS1	ATXN2	CDCA8	DIO1	DIO2	DIO3
DUOX1	DUOXA1	DUOX2	DUOXA2	EXOSC2	FGF8	FOXE1	GLIS3
GNAS	HHEX	HOXA3	IGSF1	IRS4	IYD	JAG1	KDM6A
KMT2D	NCOR2	NKX2-1	NKX2-5	NKX2-6	NTN1	P4HB	PAX8
POU1F1	PROP1	PSMA1	PSMA3	PSMD2	PSMD3	PTH1R	PTRH2
RYR2	SALL1	SECISBP2	SERPINA7	SLC16A2	SLC17A4	SLC26A4	SLC26A7
SLC30A10	SLC5A5	SLCO1C1	STAMBP	TBL1X	TBX1	TG	THRA
THRB	TPO	TRH	TRHR	TRIP11	TRIP12	TSHB	TSHR
TTR	TUBB1	UBR1	VAV3

Thyroid peroxidase (TPO) oxidizes iodide so that it can be covalently bound to tyrosine residues within TG for TH production (10). TPO consists mainly of an alpha-helical conformation with little beta sheets. Its complex structure has 3 domains: myeloperoxidase-like domain, complement control protein-like domain, and epidermal growth factor-like domain (10). These domains all play a key role in TPO for TG iodination, using hydrogen peroxide generated mainly by DUOX2, and coupling of iodinated tyrosines (11).

We present 26 Sudanese families with CH in 19, of which deleterious mutations in the functional domains of TG and TPO as well as other genes related to dyshormonogenic CH are present. Fifteen of the families are being reported for the first time, with most of the mutations being either novel or having been previously reported without a documented phenotype related to CH. Each of these mutations alters the structure and function of the molecules based on in silico modeling and its location along important functional domains of the protein. The data suggest that gene size, consanguinity, and possible positive genetic selection could account for the large proportion of dyshormonogenesis in the Sudanese population.

Materials and Methods

All patients were referred to a pediatric endocrinologist at the University of Khartoum, Sudan, presenting with stigmata of CH. Consent from patients or their guardians and their family members was obtained prior to blood sampling. Studies were approved by the University of Miami Institutional Review Board. Thyroid tests done at the time of diagnosis (thyroid-stimulating hormone [TSH] and free thyroxine [FT₄]) were done in Sudan, and subsequent serum thyroid function tests (TFTs) were completed in Miami, Florida, on the Immulite^® 1000 (Siemens, Munich, Germany) platform. TFTs included measuring levels of TSH, total thyroxine (TT₄), total triiodothyronine (TT₃), FT₄, thyroxine-binding globulin, TG, thyroid peroxidase antibodies, and thyroglobulin antibodies. Isolation of genomic deoxyribonucleic acid (DNA) from whole blood using the Qiagen QIAamp^® DNA Blood Mini Kit (Hilden, Germany) was carried out at the University of Miami. Blood samples were obtained from members of the nuclear family of each individual with CH. This included both parents and all siblings, if available. For each of the families, genomic DNA from the proband or an affected sibling along with one parent was submitted to whole-exome sequencing (WES) (Novogene, Agilent SureSelect Human All Exon V6 Kit). A compilation of thyroid genes linked to thyroid disorders shown on Table 1 was evaluated, and possible mutations linked to the phenotype were identified based on predicted functional scores, allele frequency, and zygosity. These mutations were confirmed by Sanger sequencing (Genewiz, Abi 3730xl DNA Analyzer) to verify the WES results and establish the genotype of all sampled family members along with the mode of inheritance. All identified variants were further evaluated by in silico prediction scores for how detrimental the identified variation, using Sorting Intolerant From Tolerant (SIFT) (12), Polymorphism Phenotyping version 2 Human Divergence from close mamallian homologs of human proteins (PolyPhen2 HDIV) (13), MutationTaster (14), combined annotation-dependent depletion score (CADD) (15) and the human splicing factor (16).

The variant prevalence comparison of TPO and TG genes was done using frequencies of all individuals’ whole-exome and whole-genome sequencing in the gnomAD database and whole-exome sequencing data of Sudanese samples collected in our lab along with single-nucleotide polymorphism (SNP) data of Sudanese individuals collected by Hollfelder N et al. (17). A total of 654 randomly selected SNPs were compared across the TG gene, including 152 in exons, and 234 randomly selected SNPs were also compared within the TPO gene. A paired t test was performed to compare the prevalence of mutations in individuals from all ethnic groups with Sudanese samples. For each of the 26 Sudanese families with CH, each unique causative mutation identified was tallied, and the total number of unique causative mutations for each gene was plotted against amino acid length of the gene. Differences in gender were not considered in this analysis.

Results

Gene mutations were found in 19 kindreds (Table 2) with CH (7, 18–20). In 7 families, we were unable to find the genetic cause of the CH phenotype after valuating all genes on WES with variants that were rare and predicted as deleterious by in silico prediction tools of the affected individuals. In these 7 families, we also evaluated proteins encoded by the genes on whether they interacted with the known proteins related to thyroid development, function, serum and cell TH transport, and hormone synthesis by using STRING, a database of protein–protein interactions (21). The most frequent mutations identified were in TG and TPO genes, with 7 and 6 mutations, respectively, not previously reported (Table 2). For both TG and TPO mutations, the clinical presentations were similar with goiter and/or developmental delay along with characteristic TFTs (Fig. 1). In patients with TG mutations, serum TG levels were low or undetectable (Fig. 1), averaging 1.3 ng/mL (reference range 1.7–56 ng/mL) compared with patients with TPO mutations and average TG levels of 165 ng/mL. Comprehensive clinical data and records related to the CH phenotype from many of the Sudanese families were not always readily available prior to starting medication, especially if referred from rural clinics (22).

Table 2.

Gene Variant Information and Frequencies in 19 Sudanese Families

Gene Variant Information and Frequencies
Family	Gene	Variant	CH Reported	Variant Type	RS/Accession Numbera	SIFTb	Polyphen2_HDIVc	Mutation Tasterd	CADDe	Freq_Afrf	Freq_Allg
A	DUOX1	c.C1525T, p.R509W	Yesh	MS	rs757808802	0.0, D	1.0, D	1, D	16.38	0.00	0.00
A	DUOX2	c.G3329A, p.R1110Q	Yesh	MS	rs368488511	0.003, D	0.994, D	1, D	36	0.00	0.00
B	DUOX1	c.C3388G, p.H1130D	No	MS	MN115831	0.0, D	1.0, D	1, D	23.1	.	.
B, J	DUOX2	c.1395_1396delCC, p.P465fs	Yesi	FSD	MH290757	.	.	1, D	.	.	.
C	IYD	c.C835T, p.R279C	No	MS	MN115832	0.0, D	1.0, D	1, D	25.3	.	.
D	SLC26A4	c.1197delT, p.S399fs	Yesj	FSD	rs397516413	.	.	1, D	.	0.00	0.00
E	SLC26A7	c.818_819insTTAT, p.C273fs	No	FSI	MN115835	.	.	1, D	.	.	.
F	SLC5A5	c.G749T, p.G250V	No	MS	MN115830	0.0, D	1.0, D	1, D	18.58	.	.
G	SLC5A5	c.T1042G, p.Y348D	Yesk	MS	MH046063	0.001, D	0.999, D	1, D	15.71	.	.
H, I	TPO	c.C1277G, p.A426G	No	MS	rs61758082	0.048, D	0.087, B	1, N	13.45	0.03	0.00
J	TPO	c.C1535A, p.P512H	Yesi	MS	rs150489706	0.0, D	1.0, D	1, D	25.3	0.00	0.00
J	TPO	c.G1759A, p.G587R	Yesi	MS	rs770562452	0.001, D	1.0, D	1, D	27.4	0.00	0.00
K	TPO	c.G2578A, p.G860R	No	MS	rs556552435	0.077, T	0.999, D	1, N	12	0.00	0.00
I	TPO	c.2422del, p.C808AfsTer24	No	FSD	rs763662774	.	.	1, D	23.2	0.00	0.00
L	TG	c.5975 + 401del, p.G1992GfsTer75	No	FSD	MN115834	.	.	.	.	.	.
M	TG	c.T6989A, p.V2330E	No	MS	MH137705	0.001, D	1.0, D	1, D	28.4	.	.
N	TG	c.G7021A, p.G2341S	Yesl	MS	NM003235	0.001, D	1.0, D	1, D	34	.	.
O	TG	c.7268del, p.V2423VfsTer45	No	FSD	MN115833	.	.	.	.	.	.
P	TG	c.C7655G, p.S2552X	No	SG	MH298849	.	.	1, A	46	.	.
Q	TG	c.7909ins, p.Y3637Ffs	Yesn	FSI	MH507879	.	.	.	.	.	.
Family	Gene	Variant	CH reported	Variant Type	RS/Accession number a	The Human Splicing Finder m
R, S	TG	c.4816 + 1G>T	No	SPL	MH137704	–27.7% variation: alteration of wild-type donor site, most probably affecting splicing
J	TPO	c.483–2A>G	Yesi	SPL	MH137703	–31.7% variation: alteration of wild-type donor site, most probably affecting splicing

^aRS number, reference single nucleotide variants number; Accession number, GenBank Accession Number on GenBank sequence database.

^bSorting Intolerant From Tolerant (SIFT). Scores and predictions are separated by a comma. There are 2 possible predictions: D (damaging, score ≤0.05); T (tolerated, score >0.05) (12).

^cPolymorphism Phenotyping v2 HDIV (Polyphen2_HDIV). Three possible predictions: D (probably damaging, score ≥0.909), P (possibly damaging, 0.447 ≤ score ≤ 0.908), B (benign, score ≤ 0.446) (13).

^dMutationTaster. The closer a score is to 1, the higher the confidence in the prediction. Four possible predictions: A (disease-causing automatic), D (disease-causing), N (polymorphism), P (polymorphism automatic) (14).

^eCombined Annotation Dependent Depletion (CADD) Score. Indicates rarity of variant. Score of 20 indicates variant among top 1% of deleterious variants in human genome. Score of 30 indicates variant among top 0.1% of deleterious variants in human genome (15).

^fAlternative allele frequency in African population in the Genome Aggregation Database.

^gAlternative allele frequency in all populations in the Genome Aggregation Database.

^hDual homozygous DUOX1 and DUOX2 Sudanese family previously reported by our lab (18).

ⁱCompound heterozygous TPO and DUOX2 Sudanese family previously reported in our lab (18).

^jHomozygous SLC26A4 mutation found in an affected Sudanese family, previously reported as pathogenic (19).

^kHomozygous SLC5A5 mutation in Sudanese family previously reported in our lab (20).

^lHomozygous TG mutation in Sudanese family previously reported in our lab (7).

^mThe Human Splicing Factor. Possible predictions: site broken (variation score < –10%), new site (variation score > 10%) (16).

ⁿ(27).

Abbreviations: FSD, framshift deletion; FSI, frameshift insertion mutation; MS, missense mutation; SG, stop-gainmutation; SPL, splice site mutation.

Figure 1.

Generations are denoted by roman numeral. Each subject is identified by the number just above the corresponding symbol. Laboratory thyroid function tests are aligned below each symbol. Abnormal values are in bold and underlined.

Abbreviations: FT₄, free thyroxine; TBG, thyroxine binding globulin; TG, thyroglobulin; TG Ab, anti-thyroglobulin antibody; TPO Ab, anti-TPO antibody; TSH, thyroid-stimulating hormone; TT₃, total triiodothyronine; TT₄, total thyroxine.

Mutations were also identified in the DUOX1 and DUOX2 genes, with 2 families (Table 2, family A and B) having homozygous mutations. Affected members in both family A and B presented with high TSH and low FT₄ levels prior to being diagnosed and put on treatment. One affected from each family also presented with a developmental delay due to a late diagnosis. Another family from the same Sudanese cohort (family J) also was compound heterozygous for 1 DUOX2 and 2 TPO mutations, where the affected with all 3 mutations all presented with a goiter in addition to abnormal TFTs. One mutation in each of IYD, SLC26A4, and SLC26A7 was identified in families C, D, and E, respectively (Table 2). The affected proband in each of these families also presented with high TSH and low FT₄ levels in addition to other relevant CH phenotypic considerations. All 3 affected children in family C including the proband suffered with a developmental delay and a large goiter, along with the abnormal TFTs. The affected proband in family D presented with abnormal TFTs, a goiter, developmental delay, and deafness. The proband of family E presented with abnormal TFTs as well as a developmental delay. Two families had different mutations in the SLC5A5 gene (families F and G) with all affected family members originally presenting with high TSH and low FT₄ levels and a goiter. All affected individuals except family J were homozygous for each pathogenic variant (Fig. 1 and (23)). All mutations described were linked with the phenotype of CH and were designated as deleterious by in silico prediction tools (Table 2 and (24)) and had significant alterations in protein structure (Fig. 2 and (25, 26)).

Figure 2.

Amino acid numbers are denoted by numbers spanning schematic. Important domains are denoted by roman numerals or by their name for each gene. For the TPO gene, the Myeloperoxidase (MPO) domain is denoted by the black box from amino acid position 142 to 737. Catalytic site necessary for proper TPO function is within MPO Domain. The TPO Transmembrane domain is also marked along the gene approximately between amino acid 846 and 872. Each of the documented mutations in this report and previously reported by our lab is noted by boxes in their approximate locations. *, a novel mutation. Abbreviations: CCP-L, Complement control protein-like; ChEL, Cholinesterase-like; DUOX1, Dual oxidase 1; DUOX2, Dual oxidase 2; EGF-L, Epidermal growth factor-like; IYD, Iodotyrosine Deiodinase; MPO, Myeloperoxidase; SLC26A4, Solute-carrier family 26 member 4; SLC26A7, Solute-carrier family 26 member 7; SLC5A5, Solute-carrier family 5 member 5; TG, Thyroglobulin; TPO, Thyroid peroxidase.

The mutations summarized in Table 2 directly impact the CH phenotype due to their location in the functional domains of the respective genes. Fig. 2 denotes each mutation’s approximate location along its respective gene with missense and splicing mutations occurring in highly conserved domains essential for proper protein function. Each frameshift mutation led to protein truncation.

Prevalence of TPO and TG mutations

The average prevalence of 654 SNPs in TG and 234 SNPs in TPO genes for both the Sudanese population and all other population groups are shown on Table 3. The prevalence of TG mutations was significantly higher in the Sudanese populations than in all other ethnic groups (P < 0.01). Analysis of the prevalence of mutations identified in this Sudanese cohort relative to gene length demonstrated a direct correlation of more frequent mutations seen in larger genes; the r² value (0.46694) of the trendline calculated in Excel illustrates a positive relationship between gene size and the number of unique mutations (Fig. 3).

Table 3.

T Test Results - Sudanese TG and TPO Mutation Rate vs Other Populations

		Frequency of Mutations
Gene	Number of SNPs	Sudanese (n = 234)	All (n = 123 136)	P Value
TG	654	0.1512	0.1406	≤0.01
TPO	234	0.1864	0.175	0.063

P value was calculated by t test to assess significant difference between the two average frequencies. significant P value (<0.05) is in bold. Abbreviations: SNP, single-nucleotide polymorphism; TG, thyroglobulin ; TPO, thyroid peroxidase.

Figure 3.

Relationship of unique mutations in 19 Sudanese families compared with the gene’s amino acid length. Mutations were only counted once in the instance of the same mutation being present in 2 different families. Trend line is denoted by the dotted line and the r2 is noted next to the line. Abbreviations: DUOX1, Dual oxidase 1; DUOX2, Dual oxidase 2; IYD, Iodotyrosine Deiodinase; SLC26A4, Solute-carrier family 26 member 4; SLC26A7, Solute-carrier family 26 member 7; SLC5A5, Solute-carrier family 5 member 5; TG, Thyroglobulin; TPO, Thyroid peroxidase.

Discussion

In this report, we have identified new mutations affecting 8 genes that are associated with CH. Mutations reported in DUOX1, DUOX2, IYD, SLC26A4, SLC26A7, and SLC5A5 occur along critical domains in the respective genes resulting in a clear CH phenotype in the affected individuals (Fig. 2 and (23)). Only a few mutations have been reported to date in IYD (28). While performing a radioactive iodide uptake was not possible during the family visit to the clinician in Sudan, the TFTs and clinical presentation confirm the CH diagnosis. The position of the mutation along the nitroreductase domain of IYD would hinder the ability of iodide to be properly recycled for TH synthesis, illustrated by TFT values such as markedly high TG levels (23) similar to TG values in previous families with documented IYD mutations (28). The frameshift insertion in SLC26A7 (family E) causes an early stop codon in exon 8, shortening the protein significantly from 19 exons in the wild type. Links between SLC26A7 and a CH phenotype were recently established with 10 reported cases (5, 29). The novel SLC26A7 frameshift with early truncation manifests a severe CH phenotype. The novel SLC5A5 mutation reported herein occurs along transmembrane domain 7, one of the domains essential for proper coupling and translocation of sodium (30).

Similar to previous studies of CH in different populations, TPO and TG gene mutations are the most frequent (31–34). Results from consanguineous families with CH in Turkey reported significantly high levels of TPO mutations, overshadowing all other gene mutations causing CH (31). Nonconsanguineous populations have also shown high rates of TPO mutations within such populations as the Portuguese, Japanese, and Finnish (32–34). Studies in Korea, China, and Japan also have identified DUOX2 as a frequent cause of CH (33, 35). The Sudanese population has a high degree of consanguinity (2), which may have contributed to a preponderance of a particular mutation compared with other populations. Marrying within tribes is a common practice in the Sudan (2), perpetuating a founder’s effect within each isolated tribe. However, we cannot ignore other factors that may be responsible for the high prevalence of TPO and TG gene mutations.

While missense and total mutations in TPO were found to be much higher than for other genes of comparable length (Fig. 3), the P value of 0.063 in comparing SNPs between the Sudanese population and all other populations showed no significant difference (Table 3). The damaging effects being profound, as shown in our Sudanese population, can be explained by a high prevalence of consanguinity causing a high incidence of homozygous rare mutations, as well as deficient iodine intake.

Fig. 3 illustrates the importance of considering gene length as a reason for higher rates of mutation, as our results show an increasing number of mutations with increasing gene amino acid length. The sheer size of the TG gene being more than 2700 amino acids spanning 48 exons is one of the reasons for the high incidence of TG mutations. However, other studies previously discussed (31–35) also show high rates of mutation in other genes of shorter lengths in consanguineous and nonconsanguineous families alike.

Of note, the overall incidence of SNPs in the TG gene was significantly higher in the Sudanese population compared with other reported populations (Table 3), and the possibility of positive selection was raised. As we do not have enough genetic data in the Sudanese population, positive selection in the TG gene cannot be confirmed. However, Bertranpetit and his group performed Cross Population Extended Haplotype Homozygosity (XP-EHH) test on another African population (not Sudanese), and a clear signal of recent selection in the TG gene was found (written communication). This data support the possibility of positive selection in the TG gene in the Sudanese population, although direct proof cannot be obtained at this point. Many Sudanese families live in impoverished rural areas that lack the proper intake of iodine (36). This has the potential to aggravate the severity of the clinical manifestation of CH. As previously noted, TG protein devoid of the ChEL domain would result in no TG secretion to iodination sites (37). Further work could potentially determine if TG mutations do, in fact, confer a positive selection advantage due to the potential of the goiter to retain what little iodine is available or to potentially counteract the harmful effects of high TG and thyroid differentiation from high TSH stimulation. It has been reported that the signature of positive selection was observed at some genes involved in growth and metabolism related to thyroid or pituitary function in some African populations, which may contribute to local adaptation to these mutations (38, 39). Although differences in diet, climate, and exposure to pathogens among ethnically and geographically diverse African populations are considered to produce distinct selection pressure, the mechanism is unclear. Geography might be another factor of positive selection, however further studies will be necessary to confirm this.

The rate of dyshormonogenesis in the Sudanese population is 3 times that in other populations. We find increased prevalence of TG and TPO mutations in Sudanese children with CH, and our data suggest that gene size, consanguinity, and possible positive genetic selection could account for this large proportion of dyshormonogenesis in the Sudanese population.

Glossary

Abbreviations

CH

congenital hypothyroidism

ChEL

cholinesterase-like

DNA

deoxyribonucleic acid

DUOX1

dual-oxidase 1

DUOX2

dual-oxidase 2

FT4

free thyroxine

IYD

iodotyrosine deiodinase

SLC

solute-carrier

SNP

single-nucleotide polymorphism

TFT

thyroid function test

TG

thyroglobulin

TH

thyroid hormone

TPO

thyroid peroxidase

TSH

thyroid-stimulating hormone

TT4

total thyroxine

TT3

triiodothyronine

WES

whole-exome sequencing

Additional Information

Disclosure Summary: The authors declare that they have no conflicts of interest.

Data Availability

All data generated or analyzed during this study are included in this published article or in the data repositories listed in References.